Peptide COA report, September 2026: what 600+ certificates show

A certificate of analysis is a laboratory’s statement of what it found in one sample from one lot. PepIndex reads each certificate a vendor publishes, and records five things from it: the lot identifier, the laboratory, the purity, the labeled mass and the assayed (actual) mass. A record is written only when all five can be read from the document and the document has a public URL. Certificates supplied as flat images, or with a purity figure but no net-content figure, are counted as published but not as records — the vendor gets credit for publishing them, and nothing is invented about their contents.

Every figure below is computed from the live index when this page is rendered, so it moves as vendors publish new lots and as new vendors are added. It refreshes at most once an hour. The current record set covers assays dated 6 Jul 2024 to 3 Sept 2026; 264 certificates are dated inside the last 90 days.

One thing to hold in mind throughout: all 635 records are vendor-published. None is an independently commissioned assay yet. The index scores a vendor-published certificate as a claim made in public, which is better than no claim and less than corroboration. The methodology page explains how each criterion is weighted.

Ask what separates a good research-peptide vendor from a bad one and the answer you will hear most is “purity”. The certificates do not support that. Across 635 records the median reported purity is 99.76%. The tenth percentile — the figure nine in ten certificates beat — is 99.18%. The lowest purity on record anywhere in the index is 96.1%. Only 1 certificate reports below 98%, and none below 95%.

That distribution is so tight that purity cannot rank vendors against each other. A vendor whose certificates average 99.9% and one averaging 99.3% are, on this measure, indistinguishable in any way that matters to a reader. It is also why a purity figure on its own is weak evidence: nearly everyone reports one, nearly all of them look alike, and the number says nothing about how much material is in the vial.

What purity does tell you is whether a certificate is real enough to have a method behind it. The explainer on how to read a peptide certificate of analysis covers the seven things to look for — the named HPLC method, the chromatogram, the mass-spectrometry identity check — and none of them is the percentage.

The number that varies is the assayed mass — the milligrams the laboratory found — set against the labeled mass. Half of all certificates are more than 6.9% off their label. One in ten is more than 24% off. The largest deviation on record is 109%: a vial labeled 5 mg that assayed at 10.46 mg.

The direction surprises people. Overfill is the norm, not underfill. 224 certificates show a vial more than 10% over its label; 34 show one more than 10% under. A 5 mg vial assaying at 6 mg is not a vendor being generous — it is a label that does not describe the product, and a certificate that says so in the vendor’s own document.

This is why the index scores the label discrepancy as an absolute figure. Zero deviation earns full marks on the criterion; a mean deviation of 10% or more earns nothing, in either direction. A label is a claim about quantity, and any calculation a researcher starts from that label inherits its error whichever way it runs.

56 vendors sounds like a diverse market. The testing behind it is not. Three laboratories account for 66%of every certificate record in the index. The table groups spellings of the same laboratory together — a certificate that says “Freedom Diagnostics Testing” and one that says “Freedom Diagnostics” are counted once — and shows the shortest spelling on record.

Concentration is not itself a fault — a laboratory that runs hundreds of peptide assays a month is likely to be better at it than one that runs a handful. It matters for a different reason: a certificate is only as independent as the laboratory that issued it, and when most of a market’s evidence comes from the same few sources, the accreditation and independence of those sources is the question. That is what ISO/IEC 17025 answers: it is the international standard for the competence of testing laboratories, assessed by a national accreditation body rather than asserted by the lab itself.

A pattern worth reading for on any certificate is the client field. A growing share of certificates name a company other than the storefront selling the product — the vendor is reselling stock its supplier had tested. The lot numbers still match and the test is still real, but it is the supplier’s claim rather than the vendor’s, and it tells you the vendor has not tested what it sells under its own name.

Certificates are one half of the evidence. The other half is whether a vendor can be identified and held to what it publishes. These counts come from the sourced data points on each vendor’s profile, and a fact is only counted where it is on file — an absent fact is absent, not assumed.

The gap between “publishes certificates” (35) and “names an accredited laboratory” (11) is the gap between a PDF and evidence. It is also where the vendor rankings do most of their separating: the vendors at the top are not the ones with the highest purity figures, they are the ones whose certificates can be traced to a named, checkable laboratory and matched lot by lot to what they sell.

Certificate counts follow demand. The compounds with the most lot-specific records in the index are the ones the market is stocking and turning over fastest — and the GLP-1 class now leads by a distance.

1. 01Retatrutide65 records
2. 02Tirzepatide45 records
3. 03Tesamorelin40 records
4. 04GHK-Cu38 records
5. 05BPC-15736 records
6. 06MOTS-c30 records
7. 07Ipamorelin28 records
8. 08Selank23 records
9. 09Semax23 records
10. 10TB-50022 records

Each compound page lists every vendor with a certificate on record for it, alongside the published research. The price index shows what the top-ranked vendors charge per milligram for the same compounds.

The vendors below publish the most certificates whose contents could be read into records. Volume is one input among many — the median deviation column matters at least as much — and the overall rank is shown so that neither is mistaken for the other.

Partner status never affects a rank or a place in this table; see the disclosure.

The full seven-point guide is in the certificate explainer. These are the five that the data above says matter most.

1. 01

   Find the net content line, not the purity line

   Look for "net peptide content", "assayed mass" or a per-vial milligram figure. Compare it with the label. A certificate that reports purity only has left out the number that varies most.

2. 02

   Match the lot on the certificate to the lot on the vial

   A certificate is evidence about one batch. If the lot identifier is missing, generic, or shared across many different products, it cannot be tied to what you are holding.

3. 03

   Read who the certificate was issued to

   The "client" or "submitted by" field names who sent the sample. When it is a company other than the storefront, the vendor is reselling tested stock rather than testing its own — legitimate, but a different claim.

4. 04

   Look the laboratory up

   A named lab can be checked for ISO/IEC 17025 accreditation and for a real address. "Third-party tested" without a name cannot be checked at all.

5. 05

   Check the test date

   The rubric gives full marks to certificates less than a year old and nothing to certificates two years old or more. Across this index 94% of records carry a date inside the last twelve months.